INFECTION AND IMMUNITY, Nov. 1979, p. 668-679 0019-9567/79/11-0668/10$02.00/0
Vol. 26, No. 2
Isolation and Characterization of Toxic Fractions from Brucella abortus LOUISA B. TABATABAI,* BILLY L. DEYOE, AND ALFRED E. RITCHIE National Animal Disease Center, Agricultural Research Service, Ames, Iowa 50010 Received for publication 6 August 1979
Two types of toxic fractions, protein-rich and carbohydrate-rich, were isolated from attenuated (strain 19) and virulent (strain 2308) Brucella abortus organisms. Polyacrylamide gel electrophoresis of the protein-rich fraction, in the presence and absence of sodium dodecyl sulfate, revealed qualitative and quantitative differences in the protein bands derived from the attenuated and virulent strains. Sodium dodecyl sulfate-gel electrophoresis indicated that the major differences between these protein fractions were in the molecular weight range from 14,000 to 40,000. Immunoelectrophoresis of these fractions from the attenuated and virulent strains revealed differences in the antigenic spectrum. Polypeptides in the carbohydrate-rich fraction could be visualized on polyacrylamide gels only when reacted with fluorescamine before electrophoresis. Immune sera did not precipitate the components of the carbohydrate-rich fraction. Intradermal injection of the protein and carbohydrate-rich fractions resulted in different types of skin lesions in guinea pigs, i.e., edematous/erythematous and necrotic lesions, respectively. Fractions derived from attenuated and virulent strains of B. abortus were equally toxic in the guinea pig skin test. The toxic activity of both types of fractions was susceptible to pronase and heat treatment.
Numerous reports exist of the isolation of cellular components from Brucella and their relationship to observed toxic effects in laboratory animals (3, 14, 19, 30). In most instances, cells were extracted with hot aqueous phenol as originally described by Westphal et al. (43), or cells were fractionated after total disruption with abrasives (10), with pressure (11), or by sonication (5, 23). Often Brucella fractions were less toxic in dermal hypersensitivity tests than whole cell wall or intact killed organisms (17, 30). Evidently, Brucella lipopolysaccharide (LPS) with typical endotoxic reactions is extracted mainly in the phenol phase, unlike enterobacterial LPS, which partitions into the aqueous phase (17, 24, 30, 35). Very few differences in chemical composition or toxicity have been observed in fractions obtained from smooth and rough organisms, i.e., virulent and attenuated strains of Brucella (8, 17, 24, 30). It is conceivable that by using acetone- or methanol-killed cells or by using hot aqueous phenol-extracted cells, important antigenic or potentially toxic components are extracted, denatured, or discarded. Also, these toxic or antigenic surface components might escape detection among the numerous components found in extracts (4). Our study was conducted in an attempt to demonstrate antigenic
surface components from virulent and attenuated Brucella abortus by a gentle extraction procedure and to determine their chemical composition, antigenic spectrum, and dermal toxicity. Preliminary results were presented recently (L. B. Tabatabai, B. L. Deyoe, and S. S. Stone, Fed. Proc. 37:3054, 1978). MATERIALS AND METHODS Growth of bacteria and harvesting of cultures. B. abortus was grown on tryptose agar (Difco Laboratories, Detroit, Mich.) in Roux flasks for 72 h at 37°C. Frozen stock cultures of strains 2308 (virulent) and 19 (attenuated) were maintained as a source of inoculum. Frozen material was reconstituted and passaged once on tryptose agar before Roux flasks were seeded. Cells were harvested by gentle washing of the agar surface with phosphate-buffered saline. Preparation of cell fractions. Cells were washed twice with phosphate-buffered saline and centrifuged at 20,000 X g for 20 min at 5°C. Cells were then suspended at 0.1 g/ml (wet wt/vol) in 60% methanol, stirred gently for 4 h at 5°C to inactivate the cells, and centrifuged. The aqueous methanol supernatant served as the source for the carbohydrate-rich fraction. The methanol-extracted (and inactivated) cells were the source for the protein-rich fraction. The inactivated cells were then washed twice with distilled water, suspended in 1 M NaCl-0.1 M sodium citrate at 0.2 g/ml, and stored at 5°C. The methanol inactivation-extraction procedure was repeated until culture
668
BRUCELLA ANTIGENS AND TOXIC FRACTIONS
VOL. 26, 1979
tests indicated that there were no viable cells present. Protein and carbohydrate contents of cell washes were monitored at each step. The cells were then agitated with (i) an equal volume of glass beads (0.11 mm) for 60 min at 5VC, using a Mickle tissue disintegrator (Mickle, Gomshall Surrey, England), and (ii) 30% by volume of glass beads (0.11 mm) from 2 s to 4 min under a stream of liquid carbon dioxide, using the Braun model MSK homogenizing cell (Bronwill Scientific, Inc., Rochester, N.Y.). The fractions obtained as outlined in Fig. la and lb were stored at -40'C without loss of activity.
a
669
Electron microscopy. Samples for electron microscopy were obtained from the pellet after centrifugation of the Mickle-treated cells. For negative staining, cells were added to a spot-plate well containing 1 drop of 4% potassium phosphotungstate adjusted to pH 6.8, ca. 20 drops of distilled water, and 1 drop of 1% bovine serum albumin (Cohn fraction V). After mixing gently, the dispersion was sprayed onto carboncoated, collodion-filmed grids with a Vaponefrin-type all-glass nebulizer (Ted Pella Co., Tustin, Calif.). Contrast, cell distribution, and spreading were optimized by varying the final concentrations of potassium phos-
WAASHED CELLS IN PBS
650%METHANOL.4 HR CENTIRIFUGE 20,000 xg 20 MIN SUPERNATANT FILTERED
O.45 p
PELLET WASH T|WICE WITH DISTILLED WATER CENTI'RIFUGE 20,000 zg .15 MIN
FILTER
SUPERNATANTS FILTERED
PELLET
0.45 p FILTER
FRESUSPEND
IN COLD
1M NaCI/0.1M Na
CITRATE
CENTI'RIFUGE 20, 000 X g. 15 MIN SUPERNATANT DISCARDED
PELLET 1M
FRESUSPEND IN COLD NaaCI/0.1M Na CITRATE
AGITATE IN MICKLE TISSUE DISINTEGRATER WITH 0.1 mm GLASS BEADS FOR 60 MIN, 5'C ALLOW GLASS BEADS TO SETTLE CENTRIFUGE SUPERNATANT 17, 000Xg, 30 MIN PELLET
SUPERNATANT
SAMPLED FOR ELECTRONMICROSCOPY
PRECIPITATE WITH DIALLYZE
b
(NH14)SQ4 VS.5mM
AT 70%SATURATION NH4 HCO3
WASHED CELLS IN PBS
60%METHANOL,4 HR CENTRIFUGE 20,000 xg 20 MIN SUPERNATANT 0.45p
FILTERED
PELLET
(SEE FIG.
1)
FLTER
FLASH EVAPORATE TO DRYNESS DISSOLVE RESIDUE IN 1 TO 4m1 DISTILLED WATER REPEAT TWICE DIAYLZE AGAINST 2 CHANGES OF 250m1 DISTILLED WATER, 50C FLASH EVAPORATE DIFFUSATE TO DRYNESS DISSOLVE IN 1 TO 4m1 DISTILLED WATER
FIG. 1. (a) Scheme for extracting protein-rich fraction from B. abortus. (b) Scheme for extracting carbohydrate-rich fraction from B. abortus.
670
TABATABAI, DEYOE, AND RITCHIE
photungstate, sample, or bovine serum albumin as required (31). Grids were examined immediately in a Philips EM-200 electron microscope operated at 60 kV with double-condenser illumination and a 30- to 35-t2m copper disk objective aperture. Chemical determinations. Protein content was measured by the protein-dye binding method of Bradford (6) or by the Folin-phenol method described by Lowry et al. (20), using bovine serum albumin (Armour Pharmaceutical Co., Phoenix, Ariz.) as a standard. Total carbohydrate was determined with the phenol-sulfuric acid method as described by Dubois et al. (9). Presence of 2-keto,3-deoxy sugar acids in the samples were measured as described by Ashwell (2) except that a 10-min preliminary hydrolysis step with 0.25 N H2SO4 at 1000C was included; 2-keto-3-deoxyoctonate (KDO) was used as a standard. Absorbance at 532 nm (A532) due to 2-deoxy aldoses, although minimal, was corrected for as described by Warren (38), using 2-deoxyribose as a standard. Sialic acid, also a 2-keto-3-deoxy sugar acid, was not corrected for separately, but determined enzymatically (37) on some samples. Presence of hexosamine was determined by the phenol-sulfuric acid procedure after the samples were deaminated and chromatographed according to the procedure described by Lee and Montgomery (18). Total nitrogen was determined on 100-ll samples by the micro-Kjeldahl procedure (21). Samples for amino acid analysis were hydrolyzed in vacuo for 22 h at 1100C with constantly boiling 5.7 N HCl. Samples were analyzed on a Durrum single-column instrument (Durrum Chemical Corp., Sunnyvale, Calif.) at the Department of Biochemistry and Biophysics, Iowa State University, Ames. Deoxyribonucleic acid (DNA) and ribonucleic acid were determined on 1- to 2-mg samples according to the methods of Webb and Levy (40) and Webb (39). Polyacrylamide gel electrophoresis. Sodium dodecyl sulfate (SDS)-polyacrylamide gels (7 cm long) contained 7.5% acrylamide and 0.3% bisacrylamide (41). The electrophoresis buffer used was a 1:1 dilution of the gel buffer (13), which contained 100 g of tris(hydroxymethyl)aminomethane (Tris), 10 g of ethylenediaminetetraacetic acid (disodium salt), 3.8 g of boric acid, and 1 g of SDS per 500 ml and was adjusted to pH 9.3. Electrophoresis was carried out at 7 mA/ gel at 15 to 18°C. Samples (not to exceed (100 ,l) for SDS-gel electrophoresis were prepared as described by Weber and Osborn (41). Polyacrylamide gels in the absence of SDS were prepared as described by Ortec Inc. (27), except that cylindrical gel tubes (6 mm [inner diameter] by 10 cm) were used. The gels (7 cm long) contained 7.5% acrylamide and 0.19% bisacrylamide; the stacking gel (0.5 cm long) contained 5% acrylamide and 0.13% bisacrylamide. Electrophoresis was carried out at 2.5 mA/tube at 15 to 18°C. Samples were prepared either by dialysis against Tris-borate tank buffer (27) or by addition of concentrated (fourfold) Tris-borate buffer to the proper final concentration. One drop of glycerol and 5 ,l of 0.3% bromophenyl blue were added before the samples were applied to the gels. Gels were stained for protein with Coomassie brilliant blue (41). Fluorescamine-derivatized fractions were prepared according to the method of Rosemblatt et al. (32).
INFECT. IMMUN.
Briefly, 20 ,ul of carbohydrate-rich fraction in distilled water was mixed with 20,ul of 0.1 M sodium borate in a glass tube (10 by 75 mm). To this solution was added 10 ,lO of fluorescamine (Hoffmann-LaRoche, Inc., Nutley, N.J.) at 3 mg/ml in acetone, and the mixture was allowed to stand at room temperature for 10 min. A control containing distilled water was treated in an identical manner. One drop of glycerol and 5 ,ul of 0.1% bromophenyl blue tracking dye (the latter only in control tubes) was added, and samples were loaded onto 7.5% acrylamide gels. Electrophoresis was performed at 5°C for 2 h at 2.5 mA/gel with 0.1 M Tris0.77 M glycine, pH 8.6, as the electrode buffer (7). Fluorescamine-derivatized bands were visualized under long-wave ultraviolet light. Immunoelectrophoresis. Protein fractions were concentrated with a Minicon B-15 concentrator (Amicon Corp., Lexington, Mass.) so that the samples contained 2 to 5 mg of protein per ml. Electrophoresis of 15-pl samples was carried out in 1% agarose in 0.14 M borate buffer, pH 8.6, at 24 mA/slide at room temperature. Slides were developed with hyperimmune bovine anti-B. abortus strain 19 serum and immune bovine anti-B. abortus strain 2308 serum. The hyperimmune bovine serum was prepared by multiple injections of a steer with live B. abortus strain 19. Injections were initially given intravenously and then subcutaneously or intramuscularly at biweekly intervals for 16 weeks. Organisms given at the third, fourth, and fifth injections were incorporated into Freund complete adjuvant. Sera collected were concentrated fourfold by hollow-fiber filtration (Amicon Corp.) through a 100,000-dalton filter (B. L. Deyoe, S. S. Stone, and J. B. Patterson, manuscript in preparation). The immune anti-strain 2308 serum was from an experimentally infected cow (28). Absorption of sera with whole cells. Absorption of sera with homologous B. abortus organisms was performed with 300 to 400 mg (wet weight) of methanol-inactivated cells per ml of serum. Cell suspensions were incubated for 30 min at 37°C and centrifuged. The supernatant serum was then treated two additional times with a fresh pellet of organisms. The final absorbed sera had titers of less than 20 as assayed with the standard tube test (1). Immunoelectrophoresis was then performed as described above. Toxicity assays. White female guinea pigs reared at the National Animal Disease Center, weighing 250 to 350 g, were used for the toxicity assays. The lower lateral surfaces were shaved with an Oster electric clippers equipped with a no. 40 blade. Shaved areas were then wiped with 2% tincture of iodine. Filtersterilized, concentrated, protein-rich fractions and carbohydrate-rich fractions were diluted with sterile saline to designated concentrations, and 0.1-ml quantities were injected intradermally with a 26-gauge intradermal bevel needle. Guinea pigs were observed at 1, 6, 24, and 48 h postinjection for signs of skin lesions. Representative examples of toxic reactions and control saline injection sites were photographed and examined microscopically. For histopathological examination, guinea pigs were euthanized, and then a 4-cm2 area including the injection site was excised, fixed in 10% buffered Formalin, embedded in paraffin, sectioned at 6 ,um, and stained with hematoxylin and eosin.
BRUCELLA ANTIGENS AND TOXIC FRACTIONS
VOL. 26, 1979
RESULTS
Characterization of toxic protein-rich fractions. After 60 min of shaking in the hypertonic salt solution in a Mickle apparatus, the methanol-inactivated Brucella cells were unbroken (Fig. 2a,b). Most of the cells were flattened, indicating some plasmolysis as expected. When the Braun apparatus was used at the least controllable time, 2 s, most of the cells were disrupted with massive protoplasmic extrusion (Fig. 3), precluding its use for extracting the predominantly surface-localized cell components. The ammonium sulfate-precipitated proteinrich fractions yielded approximately 1 mg of protein per mg (dry weight) of bacteria. Because of this low yield, these fractions were kept in solution rather than lyophilized. Chemical analysis of the protein-rich fraction is reported in Table 1. Carbohydrate and 2-keto-3-deoxy sugar (expressed as KDO) content varied from batch to batch. No marked difference was noted in the total carbohydrate content in the protein-rich fractions obtained from the two different strains. Total carbohydrate content (expressed as glucose) ranged from 0.05 to 0.4 mg per mg of protein. The 2-keto-3-deoxy sugar acid content (KDO) of protein-rich fractions from different batches of cells ranged from
Vol. 26, No. 2
Isolation and Characterization of Toxic Fractions from Brucella abortus LOUISA B. TABATABAI,* BILLY L. DEYOE, AND ALFRED E. RITCHIE National Animal Disease Center, Agricultural Research Service, Ames, Iowa 50010 Received for publication 6 August 1979
Two types of toxic fractions, protein-rich and carbohydrate-rich, were isolated from attenuated (strain 19) and virulent (strain 2308) Brucella abortus organisms. Polyacrylamide gel electrophoresis of the protein-rich fraction, in the presence and absence of sodium dodecyl sulfate, revealed qualitative and quantitative differences in the protein bands derived from the attenuated and virulent strains. Sodium dodecyl sulfate-gel electrophoresis indicated that the major differences between these protein fractions were in the molecular weight range from 14,000 to 40,000. Immunoelectrophoresis of these fractions from the attenuated and virulent strains revealed differences in the antigenic spectrum. Polypeptides in the carbohydrate-rich fraction could be visualized on polyacrylamide gels only when reacted with fluorescamine before electrophoresis. Immune sera did not precipitate the components of the carbohydrate-rich fraction. Intradermal injection of the protein and carbohydrate-rich fractions resulted in different types of skin lesions in guinea pigs, i.e., edematous/erythematous and necrotic lesions, respectively. Fractions derived from attenuated and virulent strains of B. abortus were equally toxic in the guinea pig skin test. The toxic activity of both types of fractions was susceptible to pronase and heat treatment.
Numerous reports exist of the isolation of cellular components from Brucella and their relationship to observed toxic effects in laboratory animals (3, 14, 19, 30). In most instances, cells were extracted with hot aqueous phenol as originally described by Westphal et al. (43), or cells were fractionated after total disruption with abrasives (10), with pressure (11), or by sonication (5, 23). Often Brucella fractions were less toxic in dermal hypersensitivity tests than whole cell wall or intact killed organisms (17, 30). Evidently, Brucella lipopolysaccharide (LPS) with typical endotoxic reactions is extracted mainly in the phenol phase, unlike enterobacterial LPS, which partitions into the aqueous phase (17, 24, 30, 35). Very few differences in chemical composition or toxicity have been observed in fractions obtained from smooth and rough organisms, i.e., virulent and attenuated strains of Brucella (8, 17, 24, 30). It is conceivable that by using acetone- or methanol-killed cells or by using hot aqueous phenol-extracted cells, important antigenic or potentially toxic components are extracted, denatured, or discarded. Also, these toxic or antigenic surface components might escape detection among the numerous components found in extracts (4). Our study was conducted in an attempt to demonstrate antigenic
surface components from virulent and attenuated Brucella abortus by a gentle extraction procedure and to determine their chemical composition, antigenic spectrum, and dermal toxicity. Preliminary results were presented recently (L. B. Tabatabai, B. L. Deyoe, and S. S. Stone, Fed. Proc. 37:3054, 1978). MATERIALS AND METHODS Growth of bacteria and harvesting of cultures. B. abortus was grown on tryptose agar (Difco Laboratories, Detroit, Mich.) in Roux flasks for 72 h at 37°C. Frozen stock cultures of strains 2308 (virulent) and 19 (attenuated) were maintained as a source of inoculum. Frozen material was reconstituted and passaged once on tryptose agar before Roux flasks were seeded. Cells were harvested by gentle washing of the agar surface with phosphate-buffered saline. Preparation of cell fractions. Cells were washed twice with phosphate-buffered saline and centrifuged at 20,000 X g for 20 min at 5°C. Cells were then suspended at 0.1 g/ml (wet wt/vol) in 60% methanol, stirred gently for 4 h at 5°C to inactivate the cells, and centrifuged. The aqueous methanol supernatant served as the source for the carbohydrate-rich fraction. The methanol-extracted (and inactivated) cells were the source for the protein-rich fraction. The inactivated cells were then washed twice with distilled water, suspended in 1 M NaCl-0.1 M sodium citrate at 0.2 g/ml, and stored at 5°C. The methanol inactivation-extraction procedure was repeated until culture
668
BRUCELLA ANTIGENS AND TOXIC FRACTIONS
VOL. 26, 1979
tests indicated that there were no viable cells present. Protein and carbohydrate contents of cell washes were monitored at each step. The cells were then agitated with (i) an equal volume of glass beads (0.11 mm) for 60 min at 5VC, using a Mickle tissue disintegrator (Mickle, Gomshall Surrey, England), and (ii) 30% by volume of glass beads (0.11 mm) from 2 s to 4 min under a stream of liquid carbon dioxide, using the Braun model MSK homogenizing cell (Bronwill Scientific, Inc., Rochester, N.Y.). The fractions obtained as outlined in Fig. la and lb were stored at -40'C without loss of activity.
a
669
Electron microscopy. Samples for electron microscopy were obtained from the pellet after centrifugation of the Mickle-treated cells. For negative staining, cells were added to a spot-plate well containing 1 drop of 4% potassium phosphotungstate adjusted to pH 6.8, ca. 20 drops of distilled water, and 1 drop of 1% bovine serum albumin (Cohn fraction V). After mixing gently, the dispersion was sprayed onto carboncoated, collodion-filmed grids with a Vaponefrin-type all-glass nebulizer (Ted Pella Co., Tustin, Calif.). Contrast, cell distribution, and spreading were optimized by varying the final concentrations of potassium phos-
WAASHED CELLS IN PBS
650%METHANOL.4 HR CENTIRIFUGE 20,000 xg 20 MIN SUPERNATANT FILTERED
O.45 p
PELLET WASH T|WICE WITH DISTILLED WATER CENTI'RIFUGE 20,000 zg .15 MIN
FILTER
SUPERNATANTS FILTERED
PELLET
0.45 p FILTER
FRESUSPEND
IN COLD
1M NaCI/0.1M Na
CITRATE
CENTI'RIFUGE 20, 000 X g. 15 MIN SUPERNATANT DISCARDED
PELLET 1M
FRESUSPEND IN COLD NaaCI/0.1M Na CITRATE
AGITATE IN MICKLE TISSUE DISINTEGRATER WITH 0.1 mm GLASS BEADS FOR 60 MIN, 5'C ALLOW GLASS BEADS TO SETTLE CENTRIFUGE SUPERNATANT 17, 000Xg, 30 MIN PELLET
SUPERNATANT
SAMPLED FOR ELECTRONMICROSCOPY
PRECIPITATE WITH DIALLYZE
b
(NH14)SQ4 VS.5mM
AT 70%SATURATION NH4 HCO3
WASHED CELLS IN PBS
60%METHANOL,4 HR CENTRIFUGE 20,000 xg 20 MIN SUPERNATANT 0.45p
FILTERED
PELLET
(SEE FIG.
1)
FLTER
FLASH EVAPORATE TO DRYNESS DISSOLVE RESIDUE IN 1 TO 4m1 DISTILLED WATER REPEAT TWICE DIAYLZE AGAINST 2 CHANGES OF 250m1 DISTILLED WATER, 50C FLASH EVAPORATE DIFFUSATE TO DRYNESS DISSOLVE IN 1 TO 4m1 DISTILLED WATER
FIG. 1. (a) Scheme for extracting protein-rich fraction from B. abortus. (b) Scheme for extracting carbohydrate-rich fraction from B. abortus.
670
TABATABAI, DEYOE, AND RITCHIE
photungstate, sample, or bovine serum albumin as required (31). Grids were examined immediately in a Philips EM-200 electron microscope operated at 60 kV with double-condenser illumination and a 30- to 35-t2m copper disk objective aperture. Chemical determinations. Protein content was measured by the protein-dye binding method of Bradford (6) or by the Folin-phenol method described by Lowry et al. (20), using bovine serum albumin (Armour Pharmaceutical Co., Phoenix, Ariz.) as a standard. Total carbohydrate was determined with the phenol-sulfuric acid method as described by Dubois et al. (9). Presence of 2-keto,3-deoxy sugar acids in the samples were measured as described by Ashwell (2) except that a 10-min preliminary hydrolysis step with 0.25 N H2SO4 at 1000C was included; 2-keto-3-deoxyoctonate (KDO) was used as a standard. Absorbance at 532 nm (A532) due to 2-deoxy aldoses, although minimal, was corrected for as described by Warren (38), using 2-deoxyribose as a standard. Sialic acid, also a 2-keto-3-deoxy sugar acid, was not corrected for separately, but determined enzymatically (37) on some samples. Presence of hexosamine was determined by the phenol-sulfuric acid procedure after the samples were deaminated and chromatographed according to the procedure described by Lee and Montgomery (18). Total nitrogen was determined on 100-ll samples by the micro-Kjeldahl procedure (21). Samples for amino acid analysis were hydrolyzed in vacuo for 22 h at 1100C with constantly boiling 5.7 N HCl. Samples were analyzed on a Durrum single-column instrument (Durrum Chemical Corp., Sunnyvale, Calif.) at the Department of Biochemistry and Biophysics, Iowa State University, Ames. Deoxyribonucleic acid (DNA) and ribonucleic acid were determined on 1- to 2-mg samples according to the methods of Webb and Levy (40) and Webb (39). Polyacrylamide gel electrophoresis. Sodium dodecyl sulfate (SDS)-polyacrylamide gels (7 cm long) contained 7.5% acrylamide and 0.3% bisacrylamide (41). The electrophoresis buffer used was a 1:1 dilution of the gel buffer (13), which contained 100 g of tris(hydroxymethyl)aminomethane (Tris), 10 g of ethylenediaminetetraacetic acid (disodium salt), 3.8 g of boric acid, and 1 g of SDS per 500 ml and was adjusted to pH 9.3. Electrophoresis was carried out at 7 mA/ gel at 15 to 18°C. Samples (not to exceed (100 ,l) for SDS-gel electrophoresis were prepared as described by Weber and Osborn (41). Polyacrylamide gels in the absence of SDS were prepared as described by Ortec Inc. (27), except that cylindrical gel tubes (6 mm [inner diameter] by 10 cm) were used. The gels (7 cm long) contained 7.5% acrylamide and 0.19% bisacrylamide; the stacking gel (0.5 cm long) contained 5% acrylamide and 0.13% bisacrylamide. Electrophoresis was carried out at 2.5 mA/tube at 15 to 18°C. Samples were prepared either by dialysis against Tris-borate tank buffer (27) or by addition of concentrated (fourfold) Tris-borate buffer to the proper final concentration. One drop of glycerol and 5 ,l of 0.3% bromophenyl blue were added before the samples were applied to the gels. Gels were stained for protein with Coomassie brilliant blue (41). Fluorescamine-derivatized fractions were prepared according to the method of Rosemblatt et al. (32).
INFECT. IMMUN.
Briefly, 20 ,ul of carbohydrate-rich fraction in distilled water was mixed with 20,ul of 0.1 M sodium borate in a glass tube (10 by 75 mm). To this solution was added 10 ,lO of fluorescamine (Hoffmann-LaRoche, Inc., Nutley, N.J.) at 3 mg/ml in acetone, and the mixture was allowed to stand at room temperature for 10 min. A control containing distilled water was treated in an identical manner. One drop of glycerol and 5 ,ul of 0.1% bromophenyl blue tracking dye (the latter only in control tubes) was added, and samples were loaded onto 7.5% acrylamide gels. Electrophoresis was performed at 5°C for 2 h at 2.5 mA/gel with 0.1 M Tris0.77 M glycine, pH 8.6, as the electrode buffer (7). Fluorescamine-derivatized bands were visualized under long-wave ultraviolet light. Immunoelectrophoresis. Protein fractions were concentrated with a Minicon B-15 concentrator (Amicon Corp., Lexington, Mass.) so that the samples contained 2 to 5 mg of protein per ml. Electrophoresis of 15-pl samples was carried out in 1% agarose in 0.14 M borate buffer, pH 8.6, at 24 mA/slide at room temperature. Slides were developed with hyperimmune bovine anti-B. abortus strain 19 serum and immune bovine anti-B. abortus strain 2308 serum. The hyperimmune bovine serum was prepared by multiple injections of a steer with live B. abortus strain 19. Injections were initially given intravenously and then subcutaneously or intramuscularly at biweekly intervals for 16 weeks. Organisms given at the third, fourth, and fifth injections were incorporated into Freund complete adjuvant. Sera collected were concentrated fourfold by hollow-fiber filtration (Amicon Corp.) through a 100,000-dalton filter (B. L. Deyoe, S. S. Stone, and J. B. Patterson, manuscript in preparation). The immune anti-strain 2308 serum was from an experimentally infected cow (28). Absorption of sera with whole cells. Absorption of sera with homologous B. abortus organisms was performed with 300 to 400 mg (wet weight) of methanol-inactivated cells per ml of serum. Cell suspensions were incubated for 30 min at 37°C and centrifuged. The supernatant serum was then treated two additional times with a fresh pellet of organisms. The final absorbed sera had titers of less than 20 as assayed with the standard tube test (1). Immunoelectrophoresis was then performed as described above. Toxicity assays. White female guinea pigs reared at the National Animal Disease Center, weighing 250 to 350 g, were used for the toxicity assays. The lower lateral surfaces were shaved with an Oster electric clippers equipped with a no. 40 blade. Shaved areas were then wiped with 2% tincture of iodine. Filtersterilized, concentrated, protein-rich fractions and carbohydrate-rich fractions were diluted with sterile saline to designated concentrations, and 0.1-ml quantities were injected intradermally with a 26-gauge intradermal bevel needle. Guinea pigs were observed at 1, 6, 24, and 48 h postinjection for signs of skin lesions. Representative examples of toxic reactions and control saline injection sites were photographed and examined microscopically. For histopathological examination, guinea pigs were euthanized, and then a 4-cm2 area including the injection site was excised, fixed in 10% buffered Formalin, embedded in paraffin, sectioned at 6 ,um, and stained with hematoxylin and eosin.
BRUCELLA ANTIGENS AND TOXIC FRACTIONS
VOL. 26, 1979
RESULTS
Characterization of toxic protein-rich fractions. After 60 min of shaking in the hypertonic salt solution in a Mickle apparatus, the methanol-inactivated Brucella cells were unbroken (Fig. 2a,b). Most of the cells were flattened, indicating some plasmolysis as expected. When the Braun apparatus was used at the least controllable time, 2 s, most of the cells were disrupted with massive protoplasmic extrusion (Fig. 3), precluding its use for extracting the predominantly surface-localized cell components. The ammonium sulfate-precipitated proteinrich fractions yielded approximately 1 mg of protein per mg (dry weight) of bacteria. Because of this low yield, these fractions were kept in solution rather than lyophilized. Chemical analysis of the protein-rich fraction is reported in Table 1. Carbohydrate and 2-keto-3-deoxy sugar (expressed as KDO) content varied from batch to batch. No marked difference was noted in the total carbohydrate content in the protein-rich fractions obtained from the two different strains. Total carbohydrate content (expressed as glucose) ranged from 0.05 to 0.4 mg per mg of protein. The 2-keto-3-deoxy sugar acid content (KDO) of protein-rich fractions from different batches of cells ranged from
